Research map / 3D vision / Human pose estimation
PoseTrack: A Benchmark for Human Pose Estimation and Tracking
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| combine MPII and PoseTrack training data instead of using MPII alone (ablation) | ArtTrack-baseline · PoseTrack test | mAP: 66.4 vs. 68.7 | Helped |
| add MPII training data to PoseTrack-only training (ablation) | ArtTrack-baseline · PoseTrack test | mAP: 55.5 vs. 68.7 | Helped |
| use MPII only rather than MPII plus PoseTrack (ablation) | ArtTrack-baseline · PoseTrack test | mAP: 68.7 vs. 66.4 | Hurt |
| raise the part detection cutoff threshold from τ=0.1 to τ=0.8 (ablation) | ArtTrack-baseline · PoseTrack test | MOTA: 38.1 to 53.4 | Helped |
| raise the part detection cutoff threshold from τ=0.1 to τ=0.8 (ablation) | ArtTrack-baseline · PoseTrack test | mAP: 68.6 -> 62.1 | Hurt |
| use a higher detection threshold to tune tracking (ablation) | ArtTrack-baseline · PoseTrack test | mAP: 68.6 -> 62.1 | Mixed |
| use the modified PAF pose model rather than PAF in ML-LAB | PoseTrack tracking · PoseTrack test | mAP: 62.5 vs. 70.3 | Helped |
| use simple frame-to-frame association on hard sequences with fast camera motion | current pose tracking approaches · PoseTrack test | MOTA | Hurt |
| evaluate on more complex, highly articulated sequences | current pose estimation and tracking methods · PoseTrack test | mAP | Hurt |
Closest papers by meaning
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · similarity 0.95)
- PoseTrack: Joint Multi-Person Pose Estimation and Tracking (CVPR 2017 · similarity 0.95)
- PoseTrack21: A Dataset for Person Search, Multi-Object Tracking and Multi-Person Pose Tracking (CVPR 2022 · similarity 0.94)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.89)
- Motions as Queries: One-Stage Multi-Person Holistic Human Motion Capture (CVPR 2025 · similarity 0.89)
- CoMotion: Concurrent Multi-person 3D Motion (ICLR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.